Identifying customers’ pain points faster. Responding in a professional and informed manner to sensitive traffic situations. Letting passengers know how to get to their destination quickly and easily. These are tasks that Deutsche Bahn customer assistance staff perform successfully every day in thousands of cases. But that can rapidly change if unforeseen incidents disrupt timetables – for example, when a train comes to a standstill because of a defect. When passengers on the platform get restless, when travel centres fill up and when train inquiries and complaints pour in by the second at Deutsche Bahn’s customer contact centre, customers are often better informed than the customer assistance staff.
The reason for this is that most rail passengers today are online when they travel. They use news available on the internet and social media platforms like Facebook, or they communicate via e-mail and WhatsApp. This means that passengers have information on the traffic situation in next to no time and exchange information.
But this information advantage is now being offset – thanks to a product initiated by DB Systel. As of June 1, “DB EarlyBird” provides a top source for many DB teams to find out about line closures, train cancellations and detours, even before customers mention them – and not only for customer-facing staff. The web application is available via Serviceportal DB Systel.
DB beats Stanford
To inform DB employees about their customers’ mobility problems as soon as possible, EarlyBird leverages data from social networks in real time. “Real Time Insights” is the new information tool’s slogan. This promise is built on a computer-assisted and algorithm-based form of speech analysis, natural language processing. Millions of data records from social media, RSS feeds and news pages are scanned using big data technologies to detect information relevant for mobility. If, for example, a user tweets about a chronically late S-Bahn in Frankfurt, this information is processed by means of text analysis for EarlyBird. It goes without saying that this is done in compliance with data protection laws. All social media data is anonymised.
“With its EarlyBird technology, DB Systel is a world leader,” says Andreas Brückner, Technical Engineer and Consultant at EarlyBird. This success is based on the big data analytics platform of the SD4M (Smart Data for Mobility) research project, which was developed by a consortium of companies led by DB Systel. It enables intelligent data management for optimising capacity utilisation, planning and forecasting in the area of mobility.
With its EarlyBird technology, DB Systel is a world leader.
The artificial intelligence-based EarlyBird application is a world-class technology because its underlying software analyses data in real time and allows up to 100 documents to be processed each second. Geolocations in Germany are recognised with an accuracy of more than 80% and when it comes to proper name recognition, EarlyBird outdoes even tools from Stanford University, the elite American institution, which is one of the world’s centres of AI research.
Watson couldn’t compete
“The motivation for developing EarlyBird was the day-to-day work of the employees at DB Dialog,” says Maren Reinsch from DB Dialog: “Frank Rennemann from DB Systel worked with us. In our discussions with employees, it became apparent that when there are unforeseen events, such as an ICE breaking down on the open track, customers are often better informed than we are. Our employees wanted a way to close that gap.”
Pretty soon, the IBM representative was the EarlyBird team’s constant companion. That was when we knew we were on the right track.
Creating the finished tool was not all plain sailing. “First, we tried using two existing systems from external companies. But that didn’t work. So we thought: okay we’ll do it ourselves,” recalls Annette Mettendorf, Product Owner of the SmartData team at DB Systel. Early Bird has no completed a number of development stages. Winning the audience prize at the Skydeck Accelerator last summer was a promising start. EarlyBird even trumped IBM, which had wanted to position its Watson AI program at Deutsche Bahn.
Expert in big data text analysis
But then work began in earnest. A team had to be set up and the platform developed and made ready for production. This was followed by a customer analysis, a business case and a billing model. The migration of around 100 million Twitter data records to the DB Enterprise Cloud has also been successfully completed. This should enable future tasks relating to storage capacity and access options to be mastered.
Andreas Brückner knows that EarlyBird is something very special: “We are a real leader in text analysis in the mobility sector. There is no system with comparable performance. Deutsche Bahn’s language is our language. So we understand it best.” That’s why the release of EarlyBird on 1 June does not mark the end of development. “We are working on various business services for our customers,” says Andreas Brückner.
The SmartData team sees the great interest in their EarlyBird service and text analytics generally as confirming the considerable potential of the technology they have developed and that many other customers could benefit from. “We are specialists in the German language,” concludes Brückner. “Our message is: If you’re analysing text data, get in touch with the SmartData team.”